Skip to content

Novel Model of Recurrent Glioblastoma May Facilitate Development of Targeted Therapies

Key findings

  • Tumor samples from a glioblastoma patient showed prominent phenotypic and genotypic changes, most notably elimination of EGFR-amplified tumor cells, after treatment with dacomitinib, a second-generation EGFR inhibitor
  • EGFR-amplified cells seemed to be only selectively eliminated by dacomitinib, such that nonamplified cells survived and seeded the re-recurrent tumor
  • These observations highlight intratumoral heterogeneity as a key mediator of tumor evasion and recurrence after targeted therapy
  • Xenografts derived from glioblastoma stem-like cell (GSC)-enriched neurospheres recapitulated the phenotypic and genotypic characteristics of the patient's tumor samples
  • In vitro, loss of EGFR amplification made GSC neurospheres less dependent on EGFR signaling for survival/proliferation, but that loss did not alter dependence on the hub pathway of PI3K-Akt

Numerous agents that target the epidermal growth factor receptor (EGFR) have been investigated for treatment of glioblastoma, based on the high prevalence of aberrant EGFR activation in the disease. Most recently, second-generation EGFR-targeted agents with irreversible inhibition and better penetration into the brain have been developed, including dacomitinib.

However, in a phase 2 trial of dacomitinib that was limited to patients with EGFR amplification, only 8% of 49 participants had a durable response. It appears that success with second-generation targeted agents will depend on identifying the escape mechanisms of glioblastoma, which might include intratumoral heterogeneity, loss of target gene expression and activation of redundant signaling pathways.

To facilitate such studies, a team led by Hiroaki Wakimoto, MD, PhD, research scientist at Massachusetts General Hospital, have created a cell model they believe will be a powerful tool for understanding how glioblastoma evolves during targeted therapy—and for screening drugs that might be efficacious against recurrent disease. Other team members included Neurosurgeons Shota Tanaka, MD, and Daniel P. Cahill, MD, PhD, and Neuro-oncologist Andrew Chi, MD, PhD, of Mass General. They describe their work, which was based on tumor samples from a single patient, in Scientific Reports.

A 62-year-old man had his glioblastoma resected and was treated with radiation and temozolomide. The tumor recurred after one year, and when it was resected a sample was used to generate a glioblastoma stem-like cell (GSC) culture.

Because the recurrent tumor harbored high-level EGFR amplification, the patient was enrolled in a trial of dacomitinib, but his cancer progressed after two months. When the re-recurrent tumor was resected, another GSC culture was generated. Despite further treatment, the patient died seven months later and an autopsy was performed.

Comparison of Tumor Samples

The researchers began by comparing samples of the original tumor, the pre-dacomitinib recurrent tumor, the post-dacomitinib re-recurrent tumor and brain obtained at autopsy. Expression of EGFR and the proliferative rate were significantly decreased in the post-dacomitinib tumor compared with the earlier samples.

The pre-dacomitinib tumor showed focal clusters of EGFR-amplified cells, suggesting that radiation and temozolomide did not preferentially target cells harboring amplified EGFR. In the post-dacomitinib tumor, by contrast, all cells had diploid EGFR signals. In the autopsy sample, there were no cells with EGFR amplification.

Loss of EGFR amplification in the post-dacomitinib tumor indicates differential efficacy of EGFR targeting depending on EGFR amplification status. EGFR-amplified cells were apparently only selectively eliminated by dacomitinib, and the non-amplified cells that survived seeded the re-recurrent tumor.

These observations highlight intratumoral heterogeneity as a key mediator of tumor evasion and recurrence after targeted therapy.

Next, the researchers established GSC-enriched neurospheres (cell clusters) that they injected into the brains of SCID mice. Like the tumor samples, the post-dacomitinib xenografts exhibited reduced levels of EGFR and a significantly reduced proliferation rate, compared with the pre-dacomitinib xenografts.

Also in concordance with the tumor samples, EGFR amplification was observed in the pre-dacomitinib xenografts but not in the post-dacomitinib xenografts.

EGFR phosphorylation was clearly reduced in post-dacomitinib GSCs, compared with pre-dacomitinib GSCs, but the phosphorylation status of Akt and Erk1/2 was comparable between the GSCs.

In Vitro Sensitivity of GSC Neurospheres to Drugs

The researchers then tested paired GSC neurospheres for their sensitivity to small-molecule drugs that target different oncogenic pathways. They assumed that the post-dacomitinib GSCs must have been derived from cells that had survived dacomitinib, and therefore they expected those GSCs to be significantly more resistant to lapatinib, a first-generation EGFR inhibitor, than pre-dacomitinib GSCs would be. That did prove to be the case.

Interestingly, both GSC lines were comparably sensitive to a PI3K inhibitor.

The findings suggest that loss of EGFR amplification made GSCs less dependent on EGFR signaling for survival/proliferation, but it did not alter dependence on PI3K-Akt, the major EGFR downstream signaling pathway.

It may be that compensatory activation of other tyrosine kinases through gain of gene amplification, and downstream signaling by acquired genetic alterations such as PTEN loss, could contribute to acquired resistance to EGFR-targeted drugs.

Whether the findings extend beyond the single patient is uncertain. If they are proven to, pairing of GSCs from before and after therapy could allow screening of drugs for recurrent glioblastoma.

Visit the Brain Tumor Stem Cell Lab

Refer a patient to the Department of Neurosurgery

Related topics


Massachusetts General Hospital neurooncologists have identified a subgroup of IDH wild-type glioblastomas that have a distinct clinical and molecular profile. The findings may have significant implications for targeted therapy in these patients.


After controlling for several potentially confounding factors, neurosurgeons at Massachusetts General Hospital found that resection of progressive glioblastoma is not significantly associated with prolonged post-progression survival, even if gross total resection is achieved.